A New Efficient Redescending M-Estimator for Robust Fitting of Linear Regression Models in the Presence of Outliers

Table 7

Total mean square error for datasets generated from (60% X & Y∼N(0, 1) and 40% X∼N(20, 1); and 20% Y∼N(10, 5)).

N	P	OLS	Huber	Hampel	Andrew	Tukey	Ali
N = 30	P = 2	5.123	5.163	0.523	0.523	0.523	0.511
	P = 3	3.703	3.725	1.264	1.283	1.278	1.212
	P = 6	2.534	2.576	2.502	2.731	2.720	2.517
N = 50	P = 2	5.044	5.061	0.257	0.258	0.258	0.207
	P = 3	3.629	3.645	0.521	0.528	0.528	0.509
	P = 6	2.534	2.417	2.127	2.159	2.157	2.012
N = 100	P = 2	5.014	5.023	0.181	0.186	0.186	0.171
	P = 3	3.576	3.585	0.229	0.229	0.229	0.219
	P = 6	2.310	2.321	1.547	1.551	1.551	1.527